traxxx/src/scrapers/tokyohot.js

'use strict';

const unprint = require('unprint');

const slugify = require('../utils/slugify');

function curateTitle(title) {
	return title?.replace(/^tokyo hot\s*/i, '');
}

function scrapeAll(scenes, channel) {
	return scenes.map(({ query }) => {
		const release = {};

		const pathname = query.url();

		release.url = unprint.prefixUrl(pathname, channel.url);
		release.entryId = pathname.match(/product\/(\w+)/)?.[1];
		release.shootId = query.attribute('img', 'title');

		release.title = curateTitle(query.content('.title'));
		release.description = query.content('.text');

		const poster = query.img();

		if (poster) {
			release.poster = [
				poster.replace('220x124', '820x462'),
				poster,
			];
		}

		return release;
	});
}

function scrapeScene({ query }, url, channel) {
	const release = {};

	release.entryId = new URL(url).pathname.match(/product\/(\w+)/)?.[1];
	release.shootId = query.content('//dt[contains(text(), "Product ID")]/following-sibling::dd[1]');

	release.title = curateTitle(query.content('.contents h2'));
	release.description = query.content('.contents .sentence');
	release.date = query.date('//dt[contains(text(), "Release Date")]/following-sibling::dd[1]', 'YYYY/MM/DD');
	release.duration = query.duration('//dt[contains(text(), "Duration")]/following-sibling::dd[1]');

	release.actors = query.all('.info a[href*="/cast"]').map((el) => ({
		name: unprint.query.content(el),
		url: unprint.query.url(el, null, { origin: channel.url }),
	}));

	release.tags = query.contents('.info a[href*="type=play"]');

	const poster = query.poster('.movie video') || query.img('.movie img');

	if (poster) {
		release.poster = [
			poster,
			poster.replace('820x462', '220x124'),
		];
	}

	release.trailer = query.video('.movie source');

	release.photos = query.imgs('.scap a', { attribute: 'href' }).map((img) => [
		img,
		img?.replace('640x480_wlimited', '150x150_default'),
	]);

	release.caps = query.imgs('.vcap a', { attribute: 'href' }).map((img) => [
		img,
		img?.replace('640x480_wlimited', '120x120_default'),
	]);

	return release;
}

// measurements are specified as a range in centimeters 85 ~ 89cm
function getMeasurement(string, inches = false) {
	if (!string) {
		return null;
	}

	const value = Array.from(string.matchAll(/(\d+(?:\.\d+)?)\s*cm/g)).at(-1)?.[1];

	if (!value) {
		return null;
	}

	if (inches) {
		return Math.round(Number(value) * 0.393701);
	}

	return Number(value);
}

function scrapeProfile({ query }) {
	const profile = {};

	const keys = query.contents('.info dt');
	const values = query.contents('.info dd');

	const bio = Object.fromEntries(keys.map((key, index) => [slugify(key, '_'), values[index]]));

	profile.birthPlace = bio.home_town;

	profile.height = getMeasurement(bio.height);

	profile.cup = bio.cup_size?.replace('cup', '').trim();
	profile.bust = getMeasurement(bio.bust_size, true);
	profile.waist = getMeasurement(bio.waist_size, true);
	profile.hip = getMeasurement(bio.hip_size || bio.hip, true);

	profile.hairStyle = bio.hair_style;
	profile.shoeSize = getMeasurement(bio.shoes_size);

	profile.bloodType = bio.blood_type?.replace('type', '').trim();

	profile.avatar = query.img('#profile img');

	return profile;
}

async function fetchLatest(channel, page) {
	const url = `${channel.url}/product/?vendor=Tokyo-Hot&page=${page}&order=published_at`;

	const res = await unprint.get(url, {
		selectAll: '#main .list .detail',
		agent: {
			rejectUnauthorized: false,
		},
	});

	if (res.ok) {
		return scrapeAll(res.context, channel);
	}

	return res.status;
}

async function fetchScene(url, channel) {
	const res = await unprint.get(url, {
		agent: {
			rejectUnauthorized: false,
		},
	});

	if (res.ok) {
		return scrapeScene(res.context, url, channel);
	}

	return res.status;
}

async function fetchProfile(actor, context) {
	if (!actor.url) {
		// search is cumbersome
		return null;
	}

	const res = await unprint.get(actor.url, {
		agent: {
			rejectUnauthorized: false,
		},
	});

	if (res.ok) {
		return scrapeProfile(res.context, context);
	}

	return res.status;
}

module.exports = {
	fetchLatest,
	fetchScene,
	fetchProfile,
};